Research On Meteorological System Of Distributed Photovoltaic Power Generation And Its Output Prediction Method

Due to the power fluctuation and intermittency of PV grid-connected power generation,PV grid-connected power generation directly affects the stable and effective source load scheduling of large-scale commercial power grids and the effective acceptance of PV power generation in large-scale commercial grids.The output quality prediction technology of photovoltaic power generation is one of the key technologies to solve this problem efficiently.In this paper,the focus of research is to study the photovoltaic system generation weather forecast related technology,in order to facilitate the development of the entire photovoltaic weather forecast system,research a variety of regional photovoltaic power generation weather forecast technology methods.In view of the basic characteristics that the short-term power generation of photovoltaic power stations is directly affected by other meteorological factors and difficult to predict,this paper proposes an automatic calculation method for the ultra-short-term power generation of photovoltaic power stations which can consider other meteorological factors.The main research contents are as follows:Through data analysis,we can clearly find that there is still a complex relationship between the output value of periodic meteorological activities of photovoltaic power station and the influence value factors of meteorological activities in different actual time periods.The correlation between meteorological characteristics and photovoltaic power generation is analyzed through the simulation of simulated data and the comparison with the historical analysis data of other experiments,which improves the accuracy and prediction of the analysis data and the analysis and calculation accuracy of the historical data model.In order to greatly improve the systematic mathematical predictability and analysis calculation precision of the mathematical prediction management model,the mathematical prediction model of short-term photovoltaic station based on Elman two dynamic neural networks is established.The results show that the optimal comprehensive prediction and analysis model constructed by the feature comprehensive selection model can achieve higher accuracy than the optimal comprehensive prediction and analysis model constructed by the feature comprehensive selection model.An optimal representational feature selection subset can be obtained from the optimal feature type selection containing conditional complementarity information,which has fewer optimal feature selection dimensions and higher unpredictable feature accuracy.